Some might say, if it’s not broken, don’t fix it, those are the same folks who fail to innovate, time and time again. In the case of CPM’s current model, it’s surely broken. Every time a CPM monte carlo is executed the process is as follows:
The user selects a duration from the range for each activity then the model runs the forward and backward pass. This yields a single iteration. It is run as many times as required in order to achieve convergence.
However there is a problem with this approach when modeling a project. It is disconnected from reality because projects actually unfold over time. If the first activity takes longer than expected, one might adjust the successor activities in order to reflect the delay in the predecessor.
In a CPM simulation this is impossible. In a Graphical Planning Method simulation we can adjust the simulation as the project unfolds. The GPM algorithm does not require a forward and backward pass and as such we can assign a value to activity (A) from the range and then apply that value before moving on to activity (B).
This is of course how we execute projects as well, in the real world. We look at the totality of progress and make adjustments accordingly. Stepping through a simulation activity by activity is a powerful modeling technique only available to those availing themselves of the GPM risk tool NetRisk.
Click here to download a copy of Dr. Gui Ponce De Leon’s “Removing the Early-Dates Bias in CPM Risk Analysis” paper.